Bond strength evaluation of heat treated Cu-Al wire bonding

Bond strength evaluation of wire bonding in microchips is the key study in any wire bonding mechanism. The quality of the wire bond interconnection relates very closely to the reliability of the microchip during performance of its function in any application. In many reports, concerns regarding the reliability of the microchip are raised due to formation of void at the wire-bond pad bonding interface, predominantly after high temperature storage (HTS) annealing conditions. In this report, the quality of wire bonds prepared at different conditions, specifically annealed at different HTS durations are determined by measurements of the strength of the interface between the bond wire and the bond pad. The samples are tested in pull test and bond shear test. It was observed that the higher bonding temperature as well as the longer duration of HTS increased the bond strength. This is represented through the analysis of the measurements of ball shear strength. This is due to the fact that higher bonding temperature and longer HTS promoted better growth of the Cu-Al IMC layer. A transmission electron microscopy - energy dispersive X-ray analysis (TEM-EDX) has been carried out to observe the formation of the Cu-Al IMC layer in the sample.

Solder Ball Robustness Comparison between SAC 387 and Polymer Solder Balls under AC and TC Reliability Test

Environment and the health concerns due to the hazardous effects of lead resulted in significant activities to find a replacement for lead-contained solders for electronics industrial. Majority of the semiconductor industrial are now replacing lead solders with Tin-Silver-Copper (SAC 387) solder balls. However, dropped balls in SAC 387 for Ball Grid Array (BGA) products due to poor solder joint strength caused by high thermal stress are a major concern in the semiconductor industries. Polymer core inside the solder ball (polymer core/Cu/Sn) is thus integrated to improve the solder ball joint strength. The function of polymer core inside the solder ball is to absorb and released the stress better as compared to the SAC 387 solder ball. Since the diffusion rate of Cu is faster than the diffusion rate of Sn, hence, this could caused the Kirkendall voids tends to form in between the Cu and Sn IMC layer. This would affect the solder ball joint strength and causing drop balls issue. By implement with an extra of Ni layer to the polymer core solder ball (core/Cu/Ni/Sn), could reduce the diffusion from Cu to Sn, thus to overcome the Kirkendall voids and to further improve the solder ball joint strength. This research work studies the performance of the solder ball shear strength of the two types of solder balls applied to MAPBGA device. In this research, both SAC 387 and polymer solder balls were went through under AC (Autoclave) and TC (Temperature Cycle) reliability test up to 144 hours and 1000 cycles, respectively. Solder ball shear strength test was conducted via Dage 4000 series bond tester. From the research work results of the two types of solder balls, the ball shear strength were decreased with an increased of aging and cycles. Overall, it can be concluded that the polymer core solder ball with an additional of Ni layer showed better performance than the polymer core without Ni layer and SAC 387 solder balls, after subjected to the AC and TC reliability test.

Reliability Study of Silver, Copper and Gold Wire Bonding on IC Device

Copper wire bonding in IC packages is not always suitable for devices with active circuit under bonding pad because higher bonding power required for copper wire bonding may cause top aluminum metal splash and mechanically impact the circuit underneath. Silver wire is an alternative solution to this problem based on its physical properties and lower cost compared to gold wire. Ag88%Au8.5%Pd2.5%X1% and Ag95%Au1.5%Pd2.5%X1% alloyed silver wires are used in the study to compare with copper and gold wires of 99.99% in purity. As bonding power plays a dominating role in wire bonding, we focused on the effects of silver, copper and gold bonding wires with different bonding power on the top aluminum metal splash of power device by Optical Microscope(OM) and Scanning Electron Microscope(SEM). The ball shear strength of the bonding wires with different bonding power in samples without mold compound encapsulation was investigated before and after 24, 48, 96 and 192 hours of pressure cooker test (PCT). The intermetallic compound (IMC) formed between silver and aluminum was confirmed by focus ion beam (FIB) and transmission electron microscope (TEM). Although the top surface of the silicon device shows no significant difference after aluminum layer removal for all three wire types, the severity level of vertical deformation and side splash of aluminum layer due to copper wire bonding is much more than silver or gold wire using same amount of bonding power. Ball shear strength of non-encapsulated silver wire decreases dramatically after PCT aging compared with copper wire or gold wire and some samples show zero shear strength after PCT 96 hours and PCT 192 hours for silver wires doped with Pd/Au. Furthermore, larger bonding power induces higher ball shear strength. The major IMC compositions between silver and aluminum are Ag3Al and Ag2Al. A thermo dynamic model was built to explain why silver wire is prone to corrosion compared with copper wire by humidity although copper is easier to be ionized than silver. No electrical test was performed as the samples cannot be tested without package encapsulation and singulation. Furthermore, silver wire samples in SO8 package with mold compound encapsulation were subjected to highly accelerated stress test (HAST), PCT, temperature cycle test (TCT) after MSL1 preconditioning test as well as high temperature operation life test (HTOL) according to JEDEC procedures. The encapsulated samples using either Ag 88wt% or Ag95wt% alloys all passed MSL1 and PCT/HAST/TCT/HTOL. Drain to source on-resistance (Rdson) of the device including package parasitics was measured and it has no significant difference between silver wire and gold wire. The results from this study shows promising data using silver alloy wires but care should be taken to further understand the degradation of silver-aluminum interface under severe humidity condition. Using other metallization on silicon top surface such as NiAu or CuAu can significantly alleviate the interface problem related to AgAl.

Effect of Structural Design Parameters on Wafer Level CSP Ball Shear Strength and Their Influence on Accelerated Thermal Cycling Reliability

Ball shear testing is typically conducted in Wafer level chip scale package (WLCSP) fabrication to estimate the strength of the solder ball attachment. Generally, the solder ball shear strength is dependent on the solder ball size, pad size, solder/pad interface treatment, reflow temperature and time. Solder ball strength is also a function of ram speed and height at which the ball is sheared with respect to the wafer. Recent investigations suggest that ball shear test is being used as an indicator for board level reliability of assemblies. In current market lead time for launching a new product is very short. Unfortunately, it takes several weeks to qualify a new product by board level qualification process. If there is a methodology through which one can predict the board level performance by extrapolating the wafer level test, it will save great amount of resources in testing and millions of dollars worth of testing time. In the first part of this study, we conducted a wafer level ball shear test. A DOE was created for varying wafer level structural parameters like solder ball size and type. Ball shear tests and Accelerated thermal cycling have similar failure signatures of compression on inner side and tension on outer side. Thus, for specific cases there is a possibility of correlating the two failure methodologies based on their failure signatures. Strain rate for ball shear test was determined based on shear speed and solder pad diameter. Strain rate for accelerated thermal cycling was determined based on difference in CTE between board and package. In this paper, results from ball shear test and accelerated thermal cycling are compared to find correlations for specific cases. The correlations derived from this study are statistical and empirical.